test: add unit tests for get_sections_for_filings and get_diffs_for_filings batch queries

[AshayK003]

Problem

db.py has two batch query functions that accept a list of filing IDs and enforce a 500-item limit via ValueError. Neither the validation path nor the actual query behavior is tested.

Functions to Test

get_sections_for_filings(conn, filing_ids) — diffiq/db.py:180

• Raises ValueError when len(filing_ids) > 500
• Returns {} for empty list
• Returns sections grouped by filing_id

get_diffs_for_filings(conn, filing_ids, stock_id) — diffiq/db.py:262

• Raises ValueError when len(filing_ids) > 500
• Returns {} for empty list
• Returns diffs grouped by filing_id_new, filtered by stock_id

Suggested Test Structure

class TestGetSectionsForFilings:
    def test_empty_list_returns_empty_dict(self, db):
        """Empty filing_ids list returns empty dict."""
        result = get_sections_for_filings(db, [])
        assert result == {}

    def test_raises_on_too_many_ids(self, db):
        """ValueError raised when filing_ids exceeds 500."""
        with pytest.raises(ValueError, match="max 500"):
            get_sections_for_filings(db, list(range(501)))

    def test_groups_sections_by_filing(self, db):
        """Sections are grouped by filing_id in the result dict."""
        stock_id = upsert_stock(db, "500295", "VEDL")
        f1 = insert_filing(db, stock_id, "u1", "2026-01-01", "F1", "https://ex.com/1.pdf")
        f2 = insert_filing(db, stock_id, "u2", "2026-01-02", "F2", "https://ex.com/2.pdf")

        insert_sections(db, f1, [
            {"header": "Opinion", "text": "v1", "chunk_hash": None, "page_num": 1, "section_idx": 0},
        ])
        insert_sections(db, f2, [
            {"header": "Opinion", "text": "v2", "chunk_hash": None, "page_num": 1, "section_idx": 0},
            {"header": "Basis", "text": "b2", "chunk_hash": None, "page_num": 2, "section_idx": 1},
        ])
        db.commit()

        result = get_sections_for_filings(db, [f1, f2])
        assert f1 in result
        assert f2 in result
        assert len(result[f1]) == 1
        assert len(result[f2]) == 2


class TestGetDiffsForFilings:
    def test_empty_list_returns_empty_dict(self, db):
        """Empty filing_ids list returns empty dict."""
        result = get_diffs_for_filings(db, [], stock_id=1)
        assert result == {}

    def test_raises_on_too_many_ids(self, db):
        """ValueError raised when filing_ids exceeds 500."""
        with pytest.raises(ValueError, match="max 500"):
            get_diffs_for_filings(db, list(range(501)), stock_id=1)

    def test_filters_by_stock_id(self, db):
        """Only diffs matching the given stock_id are returned."""
        s1 = upsert_stock(db, "500295", "VEDL")
        s2 = upsert_stock(db, "500180", "HDFCBANK")
        f1 = insert_filing(db, s1, "u1", "2026-01-01", "F1", "https://ex.com/1.pdf")
        f2 = insert_filing(db, s2, "u2", "2026-01-02", "F2", "https://ex.com/2.pdf")

        # Insert sections for FK
        db.execute("INSERT INTO sections (id, filing_id, header, text, section_idx) VALUES (1, ?, 'X', 'x', 0)", (f1,))
        db.execute("INSERT INTO sections (id, filing_id, header, text, section_idx) VALUES (2, ?, 'Y', 'y', 0)", (f2,))

        insert_diff(db, {"stock_id": s1, "filing_id_new": f1, "section_id_new": 1,
                          "section_header": "X", "diff_text": "d1", "changed": 1})
        insert_diff(db, {"stock_id": s2, "filing_id_new": f2, "section_id_new": 2,
                          "section_header": "Y", "diff_text": "d2", "changed": 1})
        db.commit()

        result = get_diffs_for_filings(db, [f1, f2], stock_id=s1)
        assert f1 in result
        assert f2 not in result  # Different stock_id

Verification

python -m pytest tests/test_db.py -v

Difficulty

Good First Issue — straightforward validation tests, uses existing fixtures and patterns from tests/test_db.py.

Labels

good first issue, tests